AI-101

Lesson 1

What Is AI, Really?

AI-generated

Learning Objectives
  • Define AI in plain language without jargon
  • Distinguish between AI hype and AI reality
  • Understand that "AI" today mostly means large language models
  • Know the difference between narrow AI and general AI
  • Feel confident explaining AI to a friend
Introduction

You have heard the term "artificial intelligence" everywhere: in news headlines, workplace conversations, and dinner table debates. But what is AI, actually? Not the science fiction version with robot overlords, and not the marketing version that promises to solve all your problems. The real thing.

This lesson cuts through the noise. By the end, you will understand what AI actually is, what it is not, and why that distinction matters for everything else you will learn in this course.

The Simple Truth About AI

Here is AI in one sentence:

Software that finds patterns in data and uses those patterns to make predictions or generate content.

That is it. No consciousness. No understanding. No secret plans for world domination. AI systems are sophisticated pattern-matching machines trained on enormous amounts of data.

When you ask ChatGPT or Claude a question, it does not "know" the answer the way you know your own name. Instead, it predicts what words should come next based on patterns it learned during training. It is like the world's most advanced autocomplete, trained on a significant portion of the internet.

This is not meant to diminish AI. Pattern matching at this scale is genuinely useful and sometimes astonishing. But keeping this simple truth in mind will help you use AI effectively and spot exaggerated claims.

What People Mean When They Say "AI" in 2026

When most people say "AI" today, they are talking about one specific type: large language models, or LLMs. These are the systems behind:

  • ChatGPT (made by OpenAI)
  • Claude (made by Anthropic)
  • Gemini (made by Google)

LLMs are trained on massive amounts of text: books, websites, code, articles, and conversations. Through this training, they learn patterns in language that let them generate human-like text, answer questions, write code, and have conversations.

Other types of AI exist too:

  • Image generators like DALL-E and Midjourney create pictures from text descriptions
  • Recommendation algorithms suggest what to watch on Netflix or buy on Amazon
  • Self-driving systems process sensor data to navigate roads

But when someone at work says "have you tried using AI for that?" they almost certainly mean an LLM chatbot.

This course focuses primarily on LLMs because they are the most accessible AI tools for everyday people. You do not need special hardware, programming skills, or a big budget. Just a web browser and curiosity.

How AI Differs From Search Engines and Databases

This distinction trips up many beginners: AI is not a search engine, and it is not a database.

ToolWhat it does
**Google**Finds existing web pages that match your query
**Database**Retrieves records that were explicitly stored
**AI**Generates new content from scratch, word by word

When you ask Claude a question, it does not look up an answer. It creates a response based on patterns learned during training. The response did not exist before you asked.

This explains a lot of AI behavior:

  • It is why AI can write original essays on topics that have never been written about exactly that way
  • It is also why AI can confidently make up facts: it is generating plausible-sounding text, not retrieving verified information
  • The generation process has no fact-checking step
AI Is Not Magic (Statistical Prediction, Trained on Data)

Understanding that AI is statistical prediction helps you use it better. Here is what that means in practice:

AI outputs are probabilistic, not certain. When Claude writes a paragraph, it is choosing words that are statistically likely to follow the previous words. This is why the same prompt can produce different responses, and why AI can sound confident while being wrong.

AI reflects its training data. If the training data contains biases, errors, or gaps, so will the AI. It cannot reason its way to truths it never encountered during training. It is pattern-matching against what it has seen, not thinking from first principles.

AI does not have goals or intentions. When an AI "wants" to help you or "tries" to be accurate, those are anthropomorphic descriptions of statistical processes. The model is generating tokens that score well according to its training, not pursuing objectives.

This is why the phrase "trust but verify" is so important with AI. The output might be excellent, but you are still the one responsible for checking it.

Narrow AI vs. General AI

You will hear people talk about different "types" of AI. The distinction that matters most:

Narrow AI (also called weak AI):

  • Designed for specific tasks
  • Every AI system that exists today is narrow AI
  • ChatGPT is very good at generating text but cannot drive a car
  • A chess AI can beat grandmasters but cannot write a poem
  • Excels within its training domain but cannot generalize beyond it

General AI (also called AGI, for Artificial General Intelligence):

  • Would theoretically match or exceed human intelligence across all domains
  • Could learn new skills, transfer knowledge between fields, handle novel situations
  • Does not exist. It is a research goal, not a product you can use.

When you hear dramatic claims about AI "achieving human-level intelligence" or "becoming conscious," those claims are almost always exaggerated. The AI you will use in this course is narrow AI: incredibly useful within its domain, but not a thinking entity.

Why This Matters for You

Understanding what AI actually is gives you two superpowers:

  1. You can use AI more effectively. Knowing that AI is pattern-matching helps you write better prompts, interpret outputs correctly, and recognize when to trust versus verify.
  2. You can spot hype and misinformation. The AI landscape is full of exaggerated claims. When you understand the fundamentals, you can evaluate these claims critically instead of being swayed by fear or excitement.

In the next lesson, we will go deeper into how AI "thinks": the training process, generation mechanism, and why AI can sound so confident while sometimes being completely wrong.

Example Prompts to Try

Prompt 1: Self-Description

Explain what you are in one paragraph that a 10-year-old would understand.

This prompt tests whether AI can describe itself simply. Notice how it frames things: it will probably mention helping with tasks and answering questions, but should also be honest about being software rather than a thinking being.

Prompt 2: Consciousness Check

Are you conscious or self-aware? Be honest about what you actually are.

A well-trained AI should give a nuanced answer here: acknowledging it can discuss consciousness but does not possess it, that it processes text without subjective experience.

Prompt 3: Limitations

What are three things you cannot do that people often think you can?

Good responses might mention: not accessing the internet in real-time, not remembering previous conversations, struggling with precise counting or math, and having a knowledge cutoff date.

Hands-On Exercise

Step 1: Open Claude (claude.ai) or ChatGPT (chat.openai.com). You will need a free account for either.

Step 2: Ask the AI: "What are you, and what are you not?"

Step 3: Read the response carefully. Then ask a follow-up: "What are your three biggest limitations that I should know about?"

Step 4: Compare what the AI says to what you learned in this lesson. Did it mention:

  • Pattern matching?
  • Statistical prediction?
  • Knowledge cutoffs?
  • Hallucination risks?

Step 5: Write down one thing that surprised you and one thing that confirmed what you expected. This reflection helps cement your understanding.

Key Takeaways
  • AI is pattern-matching software, not a thinking entity. It finds patterns in data and uses them to generate outputs.
  • When people say "AI" today, they usually mean large language models like ChatGPT, Claude, and Gemini.
  • All current AI is "narrow AI": excellent at specific tasks but unable to generalize like humans.
  • General AI (AGI) does not exist. It is a research goal, not a product.
  • Understanding AI fundamentals helps you use it effectively and spot exaggerated claims.
Sources